Where is your biotech on the data maturity ladder?

Two biotech companies, both developing promising therapeutics with strong scientific teams. In a board meeting, someone asks: "where do we stand on our IND filing?"

Company A's CSO opens her laptop and pulls up a view showing exactly which data packages are complete, which are in progress, and which are missing key information. The answer takes 30 seconds.

Company B's CSO says "let me get back to you on that" and spends the next three days coordinating with five people to piece together a complete picture from various spreadsheets, email chains, and lab notebooks.

The difference isn't talent or scientific rigor. It's where these companies sit on what we’ve come to call the data maturity ladder.

The spectrum of data maturity

Most discussions about biotech data infrastructure treat it as binary: you either “have systems” or you don’t. That framing sounds clean, but is a total oversimplification of reality. In practice, biotech data maturity is a spectrum (not a switch!). 

On one end, you have spreadsheet chaos: hundreds of files, cryptic filenames, and scientists who’ve become part-time archaeologists. On the other end, you have ML-driven digital twins: every assay, every workflow, every compound mirrored in software, feeding predictive models that make the next experiment smarter than the last.

The majority of companies live somewhere in the middle.

Each rung up the ladder changes what’s possible. You unlock new capabilities, but you also inherit new responsibilities: data governance, audit trails, user adoption, system maintenance. That’s why being honest about where you sit matters. A company preparing for an IND can’t operate like a pre-seed discovery team. But a pre-seed discovery team shouldn’t copy the infrastructure of a clinical-stage company either – it would crush them with process before they have momentum.

The goal isn’t to be perfect. The goal is to be appropriately mature for your stage, and to keep moving up deliberately instead of drifting into chaos.

Levels 1-2: Spreadsheet chaos

At the bottom of the ladder, companies work almost entirely in spreadsheets. These aren't neat, structured spreadsheets that could easily transform into a database. They're messy, with inconsistent formatting, ad-hoc columns, and different conventions across different scientists.

Teams at this level don't think in database terms. They're capturing data almost like notes on a computer, barely digital. If you ask whether they tested compound X against cell line Y, someone needs to manually search through multiple files to find out.

The hidden costs show up everywhere. Weeks get spent reconstructing experiment history for partnership discussions. Regulatory compliance becomes a constant worry because data capture isn't systematic. When the FDA asks for documentation, the scramble begins.

Levels 3-4: Structured but leaky

Companies at this level have started implementing database-like thinking. They might have an ELN, maybe some structured data capture processes. They've recognized that spreadsheet chaos doesn't scale.

But their systems are leaky. The digital representation of their R&D operations doesn't reliably mirror reality. Team members look at the system and think "I bet this isn't updated" or "I don't know if I can trust this." People still maintain shadow spreadsheets because they don't trust the official systems.

This creates a particularly frustrating situation. The company has invested in infrastructure, but the team hasn't adopted it as the source of truth. You get all the overhead of maintaining systems without the benefits of actually trusting them.

The cost shows up in duplicate work and decisions based on incomplete information. When your team doesn't treat digital systems as truth, they constantly verify through alternative channels, negating most of the efficiency gains you paid for.

Levels 5-6: Digital twin emerging

At this middle range, companies have usually achieved something critical: their team mostly trusts that digital systems represent reality. When someone updates a status in the system, others believe it and act accordingly.

The infrastructure can answer most questions, though it often requires significant manual work to pull together comprehensive views. You can generate the reports you need, but it takes time and usually involves someone who knows where everything lives.

Companies at this level can operate effectively. They meet regulatory requirements, coordinate across teams reasonably well, and make data-driven decisions. They're functional, but they're slower than competitors with more mature infrastructure.

The gap starts widening here. Companies at levels 5-6 can compete with basic operational competence, but it’s tough for them to match the speed of companies further up the ladder.

Levels 7-8: Computational maturity

These companies have fully internalized database thinking. In early conversations, they ask about API access, discuss trigger-based workflows, and want to understand computational capabilities. They've built their R&D operations around structured data from the start.

Their digital systems accurately mirror lab reality, and the entire organization acts as if maintaining this "digital twin" is part of everyone's job. When you action something in the digital system, you can trust that it will have the real-world consequences you expect.

Teams at this level often have software engineers who help build internal tools or integrations. They might have created mini custom applications that serve specific workflows unique to their science. The advantage becomes clear in decision-making speed. Questions that take Company B three days to answer take these companies 30 seconds. That compounds across hundreds of decisions over months and years.

Levels 9-10: ML-driven operations

At the top of the ladder, companies have sophisticated internal tools, automated workflows, and data pipelines that enable predictive capabilities. They’ve moved from capturing data to actually using it to inform what should happen next. 

These companies can tell you not just where every compound stands, but which ones are most likely to succeed based on historical patterns. Their systems automatically flag when something deviates from expected patterns. They've built infrastructure that makes their scientific process itself more intelligent.

The companies here have made a deliberate choice to invest in data infrastructure as a competitive advantage. They typically have small engineering teams focused entirely on R&D operations tooling.

Where do you actually sit?

Most companies overestimate their position on this ladder. Leadership thinks they're at a 6 when daily operations reveal they're closer to a 3. The gap between perception and reality creates strategic blind spots.

A few diagnostic questions can help identify where you actually are:

• Can you instantly answer where specific compounds or experiments stand, or does someone need to check and get back to you? 
• When you look at your systems, do you trust they represent reality, or do you verify through other channels? 
• If you action something in your digital systems, does your team treat that as the actual decision, or is it just documentation of a decision made elsewhere?

The trust question matters most. At lower levels, digital systems are viewed as an administrative burden, something you maintain because you have to. At higher levels, they're viewed as essential infrastructure that enables the work itself.

Where you need to be depends on your goal 

The “right” level depends entirely on your company’s goals, scale, and timeline. A pre-seed discovery team of three scientists doesn’t need an automated pipeline or a full data engineering squad. They need just enough structure to make their data findable, shareable, and reusable. Level 4 or 5 is a great place to live early on because it gives you order without killing creativity.

By contrast, a company preparing for an IND needs to be closer to level 6 or 7. Regulatory filings demand structured, auditable, and reproducible data. At that stage, “we’ll find it later” is no longer acceptable. Your ability to answer a question instantly - “which batches were used in this study?” - becomes existential.

And once you reach the clinical stage, levels 7-8 become operational requirements. Multiple programs, multiple sites, multiple teams don’t work without systems that talk to each other and a shared belief that the data is real and trustworthy.

Too often, the companies we meet panic and try to jump over multiple rungs to get to that perfect score. Our advice to them is always the same: don’t aim for a 10. Aim for the next right step. Maybe that’s moving on from spreadsheets. Maybe it’s getting people to trust your ELN. Maybe it’s finally making your “official system” the real source of truth. Whatever it is, it has to make sense for you.

--

Kaleidoscope partners with biopharma teams to help them progress up the data maturity ladder. If you're interested to chat with us and hear more on how we do that and explore whether there is a fit with your team, email us at [email protected]